Generate, summarize, adapt.
Work inside the approved workflow using the sources, tools, and limits defined for the account.
AI assurance for advertising agencies
Give the client evidence it can approve.
Petrichor turns one live account’s AI use into clear boundaries, named human decisions, current evidence, and a client-ready permission record. Leadership gets an answer it can stand behind without treating the tool as the agency’s value.
$25,000 fixed fee Four weeks One account Up to three workflows
The first conversation requires no client name or confidential material.
Why leadership needs a record
AI use is common. Accountable judgment is the distinction.
The model can produce options. It cannot hold the client relationship, defend a claim, clear a right, or accept responsibility for release. The valuable decisions still belong to people.
Select, substantiate, clear.
Reject weak work, resolve claims and rights questions, and decide what is ready for client review.
Permit, condition, prohibit.
Make the account decision against named owners, current evidence, explicit limits, and open exceptions.
Forrester and the 4As reported in June 2026 that nine in ten US marketing agencies use generative AI. WFA reported that 80% of surveyed brand owners had concerns about agency use. Adoption is common. Permission is not.
The client-facing proof
Show the boundary, the evidence, and the decision owner.
A global policy states intent. A client approval record shows what is happening in this account, who owns each consequential choice, and what would reopen the answer.
Client AI permission record
Representative structureNo client data
Decision under review
May the agency use AI-assisted creative development in this account under the conditions below?
Use position
- Permitted
- Concept exploration in approved tools using cleared, non-confidential source material.
- Conditional
- AI-involved final assets after rights review, factual review, brand approval, and a documented disclosure decision.
- Prohibited
- Restricted client data, unapproved tools, synthetic talent without cleared rights, and automated final release.
Decision evidence
- Human decisions
- Strategy, selection, claims, rights, brand, and release owners.
- Current evidence
- Tool terms, asset permissions, human contribution, review records, and open exceptions.
- Reopen triggers
- A model, supplier term, data source, partner, workflow, instruction, incident, or material exception changes.
Why Petrichor
The method is inspectable. Petrichor applies the same claim-to-evidence control to its own public claims, records limitations and review dates, and does not treat self-review as independent assurance.
Inspect the methodFee integrity
AI can shorten production. It does not erase the value of the decision.
The client is not buying keystrokes. It is buying work the agency can stand behind: a strategic choice, a usable asset, a rights position, a review record, and a person accountable for release.
Price the accountable outcome and agreed scope, not the hidden count of human minutes inside the workflow.
- Time may change
- Generation, adaptation, and production can move faster.
- Responsibility remains
- Strategy, selection, claims, rights, data, review, and release still require named owners.
- The record matters
- Assurance gives the agency and client a shared account of the work behind the fee.
Petrichor does not set agency fees or negotiate client contracts. The work makes contribution and responsibility reviewable before price is reduced to time saved by a tool.
The Account Diagnostic
One account. Three workflows. Four weeks.
Start where client scrutiny and commercial consequence meet. The Diagnostic turns one live approval question into a bounded record, a client review, and an implementation decision.
Fixed-fee engagement
$25,000
- Boundary
- One client account
- Coverage
- Up to three workflows
- Delivery
- Four weeks
Use the Diagnostic when the account has at least $100,000 of accessible annual value under a conservative 4× gate.
- 01
Account boundary
The client, SOW, workflows, tools, data, partners, dates, and exclusions.
- 02
Human decision record
Who selects, rejects, substantiates, clears, approves, and releases the work.
- 03
Rights and data evidence
Authorized inputs, licenses, restrictions, human contribution, tool terms, retention, and training treatment.
- 04
Client approval pack
Permitted, conditional, and prohibited uses, current exceptions, owners, expiry, and change triggers.
- 05
Implementation decision
A structured brand-side review, value case, funding path, and sequenced next step.
Discuss one client account
Take the decision into the room.
Boardroom deck PDFWhat can follow: multi-account implementation is scoped only after the first account establishes reusable controls, a credible funding path, and enough value for broader adoption. See the broader AI Assurance system.
Fit and reassurance
Built for a live client decision, not generic AI interest.
Strong fit
- An enterprise client has issued an AI term, restriction, questionnaire, or approval request.
- AI touches live work, final assets, audience data, or media decisions.
- A material account, pitch, renewal, or delivery model is affected.
- An agency executive owns the work and a brand-side reviewer can test the evidence.
Not yet
- The need is limited to a policy template, staff course, or tool recommendation.
- There is no live or near-term enterprise-client permission problem.
- The agency wants a badge, legal opinion, safety claim, or approval guarantee.
- The value case cannot clear the investment gate.
01What does the client receive?
A concise approval pack that states the account boundary, allowed and prohibited uses, human responsibilities, current evidence, open exceptions, expiry, and the conditions for its own permission decision.
02How is this different from an agency AI policy?
A policy states intent across the agency. AI Assurance connects one client’s conditions to the workflows, people, tools, data, partners, evidence, and decisions operating in the covered work.
03Does this certify the agency or guarantee approval?
No. The engagement is readiness and implementation work. The client owns its decision. Petrichor does not claim certification, legal compliance, safety, account retention, or guaranteed approval.
04Does AI Assurance tell us what to charge clients?
No. It makes the agency’s work, judgment, evidence, and responsibility reviewable so the commercial conversation is not reduced to hours saved by a tool. The agency and client retain their own pricing and contract decisions.
Counsel boundary
Detailed copyright and IP analysis belongs in the counsel track. The leadership record preserves the facts counsel needs without asking this page to make a legal conclusion. Open the counsel decision track.
Continue the client conversation
Bring one client question. Leave with a decision path.
We will test whether the account can be bounded, whether the evidence can support a client decision, and whether the business case earns the investment. Do not send confidential client material by email.